George explained that one of his daily ways of communicating with the company\u2019s founder was through a group chat where new software bugs or surprises were reported\u2014and he would address them. Their goals are always perfectly aligned, without need for debate: \u201cWe want to provide an L3-like intelligent driver assistance experience to all users at the current L2 BOM cost, regardless of city, driving conditions, or road type.\u201d<\/p>\n\n\n\n After the former Vice President of Autonomous Driving departed last August, George took over as the top leader of autonomous driving R&D. Within two months he completed an ambitious \u201cCrazy 200 Cities\u201d project, and in January expanded the coverage of their map-free NGP service to 243 cities nationwide. Since May this year, the founder has repeatedly emphasized that the company is the first domestic\u2014and one of only two global\u2014automobile makers to mass-produce an end-to-end large AI model (the other being Tesla). Quick to sense the crisis, the company swiftly made a series of adjustments in its organizational structure, talent recruitment, and computing resource reserves:<\/p>\n\n\n\n George revealed that the mass-produced end-to-end large model launched in May follows a \u201cone model\u201d architecture because \u201cthree networks interlace and overlap, connected by neurons rather than by interfaces defined by rules.\u201d<\/p>\n\n\n\n This week, the company is set to release its Dimensity 5.4.0 system. Compared to the version from five months ago, the new system is expected to upgrade the algorithm architecture and user experience of its smart driving features, potentially shifting the competitive landscape of China\u2019s intelligent driving industry.<\/p>\n\n\n\n Below is an excerpt from a conversation between Cross border observers: and the company\u2019s executives, discussing how autonomous driving now competes on algorithms, computing power, and data:<\/p>\n\n\n\n Cross border observers:<\/strong> George:<\/strong> Cross border observers:<\/strong> George:<\/strong> Cross border observers:<\/strong> George:<\/strong> Cross border observers:<\/strong> George:<\/strong> Additionally, our fleet is diverse\u2014from A-class to C-class vehicles, sedans to SUVs to MPVs\u2014which gives us a rich variety of high-quality data, and enables us to utilize our computing power efficiently. Although we acknowledge there is still a gap with Tesla, we are among the top domestically.<\/p>\n\n\n\n Cross border observers:<\/strong> George:<\/strong> (The dialogue continues through various topics including pricing strategies, sensor choices, data diversity, the evolution of end-to-end architectures, and strategic organizational adjustments.)<\/em><\/p>\n\n\n\n For brevity, the interview covers many additional topics\u2014from the challenges of deploying an end-to-end system on vehicle hardware to the importance of a cloud-based foundation model that can be distilled for on-vehicle use, and discussions on future Robotaxi strategies. The executives emphasize that while the shift to an end-to-end approach increases the complexity of the entire system, it is the only path to dramatically advancing from L2 to L3\/L4 autonomous driving capabilities.<\/p>\n\n\n\n They also acknowledge the challenges of ensuring that the \u201cend-to-end\u201d approach is truly integrated\u2014moving away from legacy segmented systems towards one unified model, all while maintaining explainability, safety, and efficiency. In essence, the company is betting on a future where its cloud-based foundation models, significantly larger than its current on-vehicle models, will set the ceiling for overall intelligent driving performance, and this will eventually be leveraged across their entire ecosystem of products.<\/p>\n\n\n\n <\/p>\n","protected":false},"excerpt":{"rendered":" Reinventing Cross-Border …<\/p>\n","protected":false},"author":2,"featured_media":6068,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"sfsi_plus_gutenberg_text_before_share":"","sfsi_plus_gutenberg_show_text_before_share":"","sfsi_plus_gutenberg_icon_type":"","sfsi_plus_gutenberg_icon_alignemt":"","sfsi_plus_gutenburg_max_per_row":"","footnotes":""},"categories":[27],"tags":[],"class_list":["post-6067","post","type-post","status-publish","format-standard","has-post-thumbnail","category-ai-power"],"_links":{"self":[{"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/posts\/6067","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=6067"}],"version-history":[{"count":1,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/posts\/6067\/revisions"}],"predecessor-version":[{"id":6072,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/posts\/6067\/revisions\/6072"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=\/wp\/v2\/media\/6068"}],"wp:attachment":[{"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6067"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6067"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.nexussup.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6067"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}![\"\"](\"https:\/\/blog.nexussup.com\/wp-content\/uploads\/2025\/04\/R.jpg\")
<\/figure>\n\n\n\nChanging of the Guard and New Challenges<\/h3>\n\n\n\n
However, not long afterward, even greater challenges emerged:<\/p>\n\n\n\n\n
Still, some industry experts question this claim. They note that Tesla\u2019s end-to-end solution uses one unified model for everything from perception to decision-making, whereas domestic players tend to use a \u201csegmented\u201d end-to-end model. Although the decision module replaces some rules with data-driven methods, many rules remain in the process, which some say is not as advanced.<\/p>\n\n\n\n![\"\"](\"https:\/\/blog.nexussup.com\/wp-content\/uploads\/2025\/04\/5185898b88cd98782e45d1d52ca630be.png\")
<\/figure>\n\n\n\nStrategic Adjustments and Talent Reorganization<\/h3>\n\n\n\n
\n
![\"\"](\"https:\/\/blog.nexussup.com\/wp-content\/uploads\/2025\/04\/0240.jpg_wh860.jpg\")
<\/figure>\n\n\n\nAn Exclusive Dialogue<\/h3>\n\n\n\n
Autonomous driving is starting to affect Chinese consumers\u2019 car-buying decisions. Based on your observations, how significant is this impact?<\/p>\n\n\n\n
Even if ranked outside the top three, intelligent driving is already near the top. For a vehicle priced above 300,000 RMB, lacking high-level autonomous features greatly diminishes its appeal\u2014and this trend is gradually penetrating the 200,000 RMB price segment.<\/p>\n\n\n\n
For instance, a competitor\u2019s high-end version sees over 50% adoption of its advanced driving package, with one model\u2019s option rate exceeding 70%. What are your current statistics?<\/p>\n\n\n\n
Our X9 model\u2019s high-end version currently has an option rate above 73%. Beyond that, we track comprehensive metrics\u2014activation rates of each intelligent driving feature, user penetration, and mileage penetration. In May, after launching our mass-produced end-to-end large model, the monthly user penetration for our urban smart driving service remained at roughly 83%\u201386%.<\/p>\n\n\n\n
The X9 starts at over 350,000 RMB. How are adoption rates for lower-priced models?<\/p>\n\n\n\n
Starting with our upcoming P7+ launching in early November, high-level intelligent driving will be standard across all models, eliminating the issue of an \u201coption rate.\u201d We\u2019ll deliver a software package that includes all advanced driving features, so every new model will have a 100% adoption rate.<\/p>\n\n\n\n
Standardizing high-end autonomous driving across all vehicles is a challenging feat. Many competitors can\u2019t pull that off.<\/p>\n\n\n\n
Tesla\u2019s success is partly due to its massive fleet and the large amount of training data available. While fleet size is a critical foundation, it doesn\u2019t automatically translate to quality training data\u2014only high-quality driving data is useful. It also demands strong engineering capabilities; sometimes even on-vehicle algorithms or AI are required to ensure the data is effectively utilized.<\/p>\n\n\n\n
Large amounts of data from the fleet require cloud processing. Tesla plans to invest billions in AI chips for its cloud computing, and a competitor has spent over 1 billion RMB in cloud resources, potentially reaching 1 billion USD in the future. What is your investment in cloud computing, and what is the expected growth?<\/p>\n\n\n\n
Considering our global competition, by the end of 2025 our cloud computing capacity will exceed 10 exaFLOPS\u2014an increase of 2.6 times compared to our 2024 target. However, we should not just compete on absolute computing power; factors such as quantization and model structure are also critical. We care about both the raw numbers and efficiency\u2014the utilization rate of our cloud training clusters is consistently above 80\u201390%.<\/p>\n\n\n\n![\"\"](\"https:\/\/blog.nexussup.com\/wp-content\/uploads\/2025\/04\/bf2b534c-19fe-43a8-b551-b25f39f99462.jpg\")
<\/figure>\n\n\n\n